{
  "id": 492203,
  "title": "Did anyone use EKG data?",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/492203",
  "author_name": "",
  "post_date": "2024-04-09T00:43:26.205506Z",
  "votes": 3,
  "comment_count": 6,
  "views": 0,
  "content": "<p>As above. We tried many preprocessing strategies and feature extractors, but could not make this work in our pipeline.</p>",
  "messages": [
    {
      "id": "2742538",
      "postDate": "04/09/2024 00:43:26",
      "content": "<p>As above. We tried many preprocessing strategies and feature extractors, but could not make this work in our pipeline.</p>",
      "rawMarkdown": "As above. We tried many preprocessing strategies and feature extractors, but could not make this work in our pipeline.",
      "votes": null
    },
    {
      "id": "2742541",
      "postDate": "04/09/2024 00:45:27",
      "content": "<p>I don't use EKG data. I forgot that the data also includes EKG features.</p>",
      "rawMarkdown": "I don't use EKG data. I forgot that the data also includes EKG features.",
      "votes": null
    },
    {
      "id": "2742542",
      "postDate": "04/09/2024 00:45:27",
      "content": "<p>I did for completeness (tried to use as much data as possible), but it didn't increase or decrease my CV or LB. </p>",
      "rawMarkdown": "I did for completeness (tried to use as much data as possible), but it didn't increase or decrease my CV or LB.",
      "votes": null
    },
    {
      "id": "2742555",
      "postDate": "04/09/2024 00:52:37",
      "content": "<p>I used it for some models but it could not help much I think.</p>",
      "rawMarkdown": "I used it for some models but it could not help much I think.",
      "votes": null
    },
    {
      "id": "2742742",
      "postDate": "04/09/2024 03:52:35",
      "content": "<p>EKG was only usable in 1D models if we normalize each channel separately. Otherwise, it was messing out 2D inputs because of its extreme values. We didn't use it at the end.</p>",
      "rawMarkdown": "EKG was only usable in 1D models if we normalize each channel separately. Otherwise, it was messing out 2D inputs because of its extreme values. We didn't use it at the end.",
      "votes": null
    },
    {
      "id": "2743084",
      "postDate": "04/09/2024 08:08:34",
      "content": "<p>We tried to use it in 1D like Gunes pointed out but it did not help the CV / LB.</p>\n<p>Best,<br>\nJan</p>",
      "rawMarkdown": "We tried to use it in 1D like Gunes pointed out but it did not help the CV / LB.\n\nBest,\nJan",
      "votes": null
    },
    {
      "id": "2744137",
      "postDate": "04/09/2024 19:18:16",
      "content": "<p>I tried using various techniques but it did not improve my CV score. (And, I did not submit to LB so I don't know the effect on LB score).</p>",
      "rawMarkdown": "I tried using various techniques but it did not improve my CV score. (And, I did not submit to LB so I don't know the effect on LB score).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2742541,
      "author_name": "quan0095",
      "author_url": "",
      "post_date": "04/09/2024 00:45:27",
      "content": "<p>I don't use EKG data. I forgot that the data also includes EKG features.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2742542,
      "author_name": "ryanirl",
      "author_url": "",
      "post_date": "04/09/2024 00:45:27",
      "content": "<p>I did for completeness (tried to use as much data as possible), but it didn't increase or decrease my CV or LB. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2742555,
      "author_name": "goldenlock",
      "author_url": "",
      "post_date": "04/09/2024 00:52:37",
      "content": "<p>I used it for some models but it could not help much I think.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2742742,
      "author_name": "gunesevitan",
      "author_url": "",
      "post_date": "04/09/2024 03:52:35",
      "content": "<p>EKG was only usable in 1D models if we normalize each channel separately. Otherwise, it was messing out 2D inputs because of its extreme values. We didn't use it at the end.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2743084,
      "author_name": "janbrederecke",
      "author_url": "",
      "post_date": "04/09/2024 08:08:34",
      "content": "<p>We tried to use it in 1D like Gunes pointed out but it did not help the CV / LB.</p>\n<p>Best,<br>\nJan</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2744137,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "04/09/2024 19:18:16",
      "content": "<p>I tried using various techniques but it did not improve my CV score. (And, I did not submit to LB so I don't know the effect on LB score).</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2742538": "As above. We tried many preprocessing strategies and feature extractors, but could not make this work in our pipeline.",
    "2742541": "I don't use EKG data. I forgot that the data also includes EKG features.",
    "2742542": "I did for completeness (tried to use as much data as possible), but it didn't increase or decrease my CV or LB.",
    "2742555": "I used it for some models but it could not help much I think.",
    "2742742": "EKG was only usable in 1D models if we normalize each channel separately. Otherwise, it was messing out 2D inputs because of its extreme values. We didn't use it at the end.",
    "2743084": "We tried to use it in 1D like Gunes pointed out but it did not help the CV / LB.\n\nBest,\nJan",
    "2744137": "I tried using various techniques but it did not improve my CV score. (And, I did not submit to LB so I don't know the effect on LB score)."
  },
  "source": "meta"
}